In this paper the capabilities of an incoherent radar sensor network for robust Doppler-based gesture recognition are investigated and a significant performance boost is demonstrated. A comprehensive dataset is recorded with an incoherent sensor network consisting of three time-synchronized 77 GHz FMCW radars. Based on this dataset, we show that differential Doppler features obtained from the varying viewing angles result in a significant multistatic gain for classification, particularly for high intra-class variations and low Doppler frequencies. For the most complex dataset crossuser validation accuracy of a CNN with optimized data fusion is improved by 7.4 % to an overall value of 87.1 %, which we regard high as gestures are not designed for distinguishability but reflect everyday control and communication signals.
Radar-based gesture recognition can provide autonomous electronic systems with a reliable way to infer a human's intention, e.g. in traffic environments involving vulnerable road users (VRUs). Particularly in complex scenarios, algorithms operating on radar target lists derived from constant false-alarm rate (CFAR) outputs present an attractive solution, as they not only enable the filtering of relevant targets but can also make full use of the diverse, high-resolution target parameters provided by modern radar sensors. Therefore, this paper proposes PointNet+LSTM for the enhanced target list-based recognition of challenging traffic gestures, combining per-frame feature extraction with PointNet and learning from sequences with a long short-term memory (LSTM). The approach is generalized to facilitate the use of multistatic radar data from sensor networks to exploit slightly different viewing angles, which is particularly helpful for motions with low radial velocity. The proposed method is validated on a comprehensive dataset comprising eight traffic gestures and data recorded from 35 participants. Measurements are conducted both indoors and outdoors with an incoherent radar sensor network comprising three chirp sequence (CS)-multiple-input multiple-output (MIMO) sensors. On this challenging dataset, our approach clearly outperforms a reference convolutional neural network (CNN), reaching up to 92.2 % cross-validation accuracy.
<p>Massive
multiple-input multiple-output (MIMO) systems operating in the centimeter-wave
(cmWave) and millimeter-wave (mmWave) region offer huge spectral efficiencies,
which enable to satisfy the urgent need for higher data rates in mobile
communication networks. However, the proper design of those massive MIMO
systems first requires a deep understanding of the underlying wireless
propagation channel. Therefore, we present a fully-digital MIMO measurement
system operating around 28 GHz. The system enables to take fast subsequent
snapshots of the complex MIMO channel matrix. Based on this method we
statistically analyze the time-dependent channel behavior, the achievable
signal quality and spectral efficiency, as well as the channel eigenvalue
profile. Furthermore, the presented calibration approach for the receiver
enables an estimation of the dominant absolute angle of arrival (AoA) and
allows us to draw conclusions about the line-of-sight (LOS) dominance of the
scenario. In total, 159 uplink communication measurements over 20 seconds are
conducted in three different small cell site scenarios to investigate the
wireless propagation behavior. The measurements reveal the existence of several
spatial propagation paths between the mobile transmitter and the base station.
Furthermore, an insight into their likelihood in different propagation
scenarios is also given.</p>
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.